1<?php 2/** 3 * DokuWiki indexer 4 * 5 * @license GPL 2 (http://www.gnu.org/licenses/gpl.html) 6 * @author Andreas Gohr <andi@splitbrain.org> 7 */ 8if(!defined('DOKU_INC')) define('DOKU_INC',dirname(__FILE__).'/../../'); 9define('DOKU_DISABLE_GZIP_OUTPUT',1); 10require_once(DOKU_INC.'inc/init.php'); 11session_write_close(); //close session 12if(!defined('NL')) define('NL',"\n"); 13 14// Version tag used to force rebuild on upgrade 15define('INDEXER_VERSION', 2); 16 17// keep running after browser closes connection 18@ignore_user_abort(true); 19 20// check if user abort worked, if yes send output early 21$defer = !@ignore_user_abort() || $conf['broken_iua']; 22if(!$defer){ 23 sendGIF(); // send gif 24} 25 26$ID = cleanID($_REQUEST['id']); 27 28// Catch any possible output (e.g. errors) 29$output = isset($_REQUEST['debug']) && $conf['allowdebug']; 30if(!$output) ob_start(); 31 32// run one of the jobs 33$tmp = array(); // No event data 34$evt = new Doku_Event('INDEXER_TASKS_RUN', $tmp); 35if ($evt->advise_before()) { 36 runIndexer() or 37 metaUpdate() or 38 runSitemapper() or 39 sendDigest() or 40 runTrimRecentChanges() or 41 runTrimRecentChanges(true) or 42 $evt->advise_after(); 43} 44if($defer) sendGIF(); 45 46if(!$output) ob_end_clean(); 47exit; 48 49// -------------------------------------------------------------------- 50 51/** 52 * Trims the recent changes cache (or imports the old changelog) as needed. 53 * 54 * @param media_changes If the media changelog shall be trimmed instead of 55 * the page changelog 56 * 57 * @author Ben Coburn <btcoburn@silicodon.net> 58 */ 59function runTrimRecentChanges($media_changes = false) { 60 global $conf; 61 62 $fn = ($media_changes ? $conf['media_changelog'] : $conf['changelog']); 63 64 // Trim the Recent Changes 65 // Trims the recent changes cache to the last $conf['changes_days'] recent 66 // changes or $conf['recent'] items, which ever is larger. 67 // The trimming is only done once a day. 68 if (@file_exists($fn) && 69 (@filemtime($fn.'.trimmed')+86400)<time() && 70 !@file_exists($fn.'_tmp')) { 71 @touch($fn.'.trimmed'); 72 io_lock($fn); 73 $lines = file($fn); 74 if (count($lines)<=$conf['recent']) { 75 // nothing to trim 76 io_unlock($fn); 77 return false; 78 } 79 80 io_saveFile($fn.'_tmp', ''); // presave tmp as 2nd lock 81 $trim_time = time() - $conf['recent_days']*86400; 82 $out_lines = array(); 83 84 for ($i=0; $i<count($lines); $i++) { 85 $log = parseChangelogLine($lines[$i]); 86 if ($log === false) continue; // discard junk 87 if ($log['date'] < $trim_time) { 88 $old_lines[$log['date'].".$i"] = $lines[$i]; // keep old lines for now (append .$i to prevent key collisions) 89 } else { 90 $out_lines[$log['date'].".$i"] = $lines[$i]; // definitely keep these lines 91 } 92 } 93 94 if (count($lines)==count($out_lines)) { 95 // nothing to trim 96 @unlink($fn.'_tmp'); 97 io_unlock($fn); 98 return false; 99 } 100 101 // sort the final result, it shouldn't be necessary, 102 // however the extra robustness in making the changelog cache self-correcting is worth it 103 ksort($out_lines); 104 $extra = $conf['recent'] - count($out_lines); // do we need extra lines do bring us up to minimum 105 if ($extra > 0) { 106 ksort($old_lines); 107 $out_lines = array_merge(array_slice($old_lines,-$extra),$out_lines); 108 } 109 110 // save trimmed changelog 111 io_saveFile($fn.'_tmp', implode('', $out_lines)); 112 @unlink($fn); 113 if (!rename($fn.'_tmp', $fn)) { 114 // rename failed so try another way... 115 io_unlock($fn); 116 io_saveFile($fn, implode('', $out_lines)); 117 @unlink($fn.'_tmp'); 118 } else { 119 io_unlock($fn); 120 } 121 return true; 122 } 123 124 // nothing done 125 return false; 126} 127 128/** 129 * Runs the indexer for the current page 130 * 131 * @author Andreas Gohr <andi@splitbrain.org> 132 */ 133function runIndexer(){ 134 global $ID; 135 global $conf; 136 print "runIndexer(): started".NL; 137 138 if(!$ID) return false; 139 140 // check if indexing needed 141 $idxtag = metaFN($ID,'.indexed'); 142 if(@file_exists($idxtag)){ 143 if(trim(io_readFile($idxtag)) == INDEXER_VERSION){ 144 $last = @filemtime($idxtag); 145 if($last > @filemtime(wikiFN($ID))){ 146 print "runIndexer(): index for $ID up to date".NL; 147 return false; 148 } 149 } 150 } 151 152 // try to aquire a lock 153 $lock = $conf['lockdir'].'/_indexer.lock'; 154 while(!@mkdir($lock,$conf['dmode'])){ 155 usleep(50); 156 if(time()-@filemtime($lock) > 60*5){ 157 // looks like a stale lock - remove it 158 @rmdir($lock); 159 print "runIndexer(): stale lock removed".NL; 160 }else{ 161 print "runIndexer(): indexer locked".NL; 162 return false; 163 } 164 } 165 if($conf['dperm']) chmod($lock, $conf['dperm']); 166 167 // do the work 168 idx_addPage($ID); 169 170 // we're finished - save and free lock 171 io_saveFile(metaFN($ID,'.indexed'),INDEXER_VERSION); 172 @rmdir($lock); 173 print "runIndexer(): finished".NL; 174 return true; 175} 176 177/** 178 * Will render the metadata for the page if not exists yet 179 * 180 * This makes sure pages which are created from outside DokuWiki will 181 * gain their data when viewed for the first time. 182 */ 183function metaUpdate(){ 184 global $ID; 185 print "metaUpdate(): started".NL; 186 187 if(!$ID) return false; 188 $file = metaFN($ID, '.meta'); 189 echo "meta file: $file".NL; 190 191 // rendering needed? 192 if (@file_exists($file)) return false; 193 if (!page_exists($ID)) return false; 194 195 global $conf; 196 197 // gather some additional info from changelog 198 $info = io_grep($conf['changelog'], 199 '/^(\d+)\t(\d+\.\d+\.\d+\.\d+)\t'.preg_quote($ID,'/').'\t([^\t]+)\t([^\t\n]+)/', 200 0,true); 201 202 $meta = array(); 203 if(!empty($info)){ 204 $meta['date']['created'] = $info[0][1]; 205 foreach($info as $item){ 206 if($item[4] != '*'){ 207 $meta['date']['modified'] = $item[1]; 208 if($item[3]){ 209 $meta['contributor'][$item[3]] = $item[3]; 210 } 211 } 212 } 213 } 214 215 $meta = p_render_metadata($ID, $meta); 216 p_save_metadata($ID, $meta); 217 218 echo "metaUpdate(): finished".NL; 219 return true; 220} 221 222/** 223 * Builds a Google Sitemap of all public pages known to the indexer 224 * 225 * The map is placed in the root directory named sitemap.xml.gz - This 226 * file needs to be writable! 227 * 228 * @author Andreas Gohr 229 * @link https://www.google.com/webmasters/sitemaps/docs/en/about.html 230 */ 231function runSitemapper(){ 232 print "runSitemapper(): started".NL; 233 $result = Sitemapper::generate() && Sitemapper::pingSearchEngines(); 234 print 'runSitemapper(): finished'.NL; 235 return $result; 236} 237 238/** 239 * Send digest and list mails for all subscriptions which are in effect for the 240 * current page 241 * 242 * @author Adrian Lang <lang@cosmocode.de> 243 */ 244function sendDigest() { 245 echo 'sendDigest(): start'.NL; 246 global $ID; 247 global $conf; 248 if (!$conf['subscribers']) { 249 return; 250 } 251 $subscriptions = subscription_find($ID, array('style' => '(digest|list)', 252 'escaped' => true)); 253 global $auth; 254 global $lang; 255 global $conf; 256 global $USERINFO; 257 258 // remember current user info 259 $olduinfo = $USERINFO; 260 $olduser = $_SERVER['REMOTE_USER']; 261 262 foreach($subscriptions as $id => $users) { 263 if (!subscription_lock($id)) { 264 continue; 265 } 266 foreach($users as $data) { 267 list($user, $style, $lastupdate) = $data; 268 $lastupdate = (int) $lastupdate; 269 if ($lastupdate + $conf['subscribe_time'] > time()) { 270 // Less than the configured time period passed since last 271 // update. 272 continue; 273 } 274 275 // Work as the user to make sure ACLs apply correctly 276 $USERINFO = $auth->getUserData($user); 277 $_SERVER['REMOTE_USER'] = $user; 278 if ($USERINFO === false) { 279 continue; 280 } 281 282 if (substr($id, -1, 1) === ':') { 283 // The subscription target is a namespace 284 $changes = getRecentsSince($lastupdate, null, getNS($id)); 285 } else { 286 if(auth_quickaclcheck($id) < AUTH_READ) continue; 287 288 $meta = p_get_metadata($id); 289 $changes = array($meta['last_change']); 290 } 291 292 // Filter out pages only changed in small and own edits 293 $change_ids = array(); 294 foreach($changes as $rev) { 295 $n = 0; 296 while (!is_null($rev) && $rev['date'] >= $lastupdate && 297 ($_SERVER['REMOTE_USER'] === $rev['user'] || 298 $rev['type'] === DOKU_CHANGE_TYPE_MINOR_EDIT)) { 299 $rev = getRevisions($rev['id'], $n++, 1); 300 $rev = (count($rev) > 0) ? $rev[0] : null; 301 } 302 303 if (!is_null($rev) && $rev['date'] >= $lastupdate) { 304 // Some change was not a minor one and not by myself 305 $change_ids[] = $rev['id']; 306 } 307 } 308 309 if ($style === 'digest') { 310 foreach($change_ids as $change_id) { 311 subscription_send_digest($USERINFO['mail'], $change_id, 312 $lastupdate); 313 } 314 } elseif ($style === 'list') { 315 subscription_send_list($USERINFO['mail'], $change_ids, $id); 316 } 317 // TODO: Handle duplicate subscriptions. 318 319 // Update notification time. 320 subscription_set($user, $id, $style, time(), true); 321 } 322 subscription_unlock($id); 323 } 324 325 // restore current user info 326 $USERINFO = $olduinfo; 327 $_SERVER['REMOTE_USER'] = $olduser; 328} 329 330/** 331 * Just send a 1x1 pixel blank gif to the browser 332 * 333 * @author Andreas Gohr <andi@splitbrain.org> 334 * @author Harry Fuecks <fuecks@gmail.com> 335 */ 336function sendGIF(){ 337 if(isset($_REQUEST['debug'])){ 338 header('Content-Type: text/plain'); 339 return; 340 } 341 $img = base64_decode('R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAEALAAAAAABAAEAAAIBTAA7'); 342 header('Content-Type: image/gif'); 343 header('Content-Length: '.strlen($img)); 344 header('Connection: Close'); 345 print $img; 346 flush(); 347 // Browser should drop connection after this 348 // Thinks it's got the whole image 349} 350 351//Setup VIM: ex: et ts=4 enc=utf-8 : 352// No trailing PHP closing tag - no output please! 353// See Note at http://www.php.net/manual/en/language.basic-syntax.instruction-separation.php 354